Google

During the second stage of a multi-round interaction, a prompt template is used to instruct a Large Language Model to assess and improve a previously generated solution. This template presents both the original question and the initial response to the model. An example prompt is: "You will be provided with a math problem, along with a solution. Evaluate the correctness of this solution, and identify any errors if present. Then, work out your own solution. Problem: $${*\mathrm{problem}*}$$ Solution: $${*\mathrm{solution}*}$$".

Prompt Template for Answer Evaluation and Refinement

A developer is using the following prompt to have a language model check a student's math homework:

`Problem: What is 5 * (3 + 2)?
Student's Answer: 25.
Is the student's answer correct? If not, what is the right answer?`

Analyze this prompt. Identify two key weaknesses in its structure for reliably guiding the model to both evaluate the student's work and provide a high-quality correction. Then, rewrite the prompt to be more effective.

Improving a Flawed Evaluation Prompt

Based on the case study below, design a single, more effective prompt for the developer's task. Explain the key components of your proposed prompt and why it represents an improvement over the original approach.

Automating Homework Grading with an LLM

A developer is designing a two-step process for a language model to solve complex logic puzzles. In the first step, the model provides an initial solution. For the second step, the developer wants to prompt the model to critically review and improve its own initial answer. Which of the following prompts for the second step is best designed to encourage self-correction and produce a more accurate final solution?

Learn Before

Related